2036 OPTICS LETTERS / Vol. 20, No. 19 / October 1, 1995

Detection of methyl radicals in a flat flame bydegenerate four-wave mixing

Volker Sick,* Mary N. Bui-Pham, and Roger L. Farrow

Combustion Research Facility, MS 9057, Sandia National Laboratories, Livermore, California 94551

Received March 28, 1995

We report the spatially resolved detection of methyl radicals in a methane–air f lat f lame, using degeneratefour-wave mixing (DFWM). A frequency-tripled dye laser pumped with a frequency-doubled Nd:YAG laser wasused to access the Herzberg b1 band of methyl near 216 nm. Using a nearly phase-conjugate geometry, wedetected methyl with high spatial resolution [0.2 mm (0.3 mm) vertical (horizontal) and ,6 mm longitudinal]and with good signal-to-noise ratio in a rich sf ­ 1.55d f lame. Compared with laser absorption spectra,DFWM spectra were much less inf luenced by a broad featureless background. From the absorption data, wemeasured the peak methyl concentration to be 650 parts in 106, resulting in an estimated DFWM detectionlimit of 65 parts in 106. 1995 Optical Society of America

Owing to the importance of the methyl radicalsCH3d in combustion and in the growth of diamond bychemical-vapor deposition,1,2 the development of opticaltechniques for measuring CH3 concentrations underhigh-temperature conditions and nonuniform spatialdistribution has been widely pursued. Since thefirst demonstration of the existence of accessibleUV transitions in CH3,3 UV absorption methods,4 – 6

including cavity ring-down spectroscopy,7 have beenwidely used for line-of-sight measurements. Tech-niques with improved spatial precision, i.e., resolutionin all three dimensions, have since been developed.Resonance-enhanced multiphoton ionization hasbeen demonstrated for sensitive detection of CH3 inf lames,8 and Raman9 and coherent Raman anti-Stokesspectroscopy10 have been used to study CH3 at muchhigher concentrations. Laser-induced f luorescence,commonly used for many combustion-relevant inter-mediate species, cannot be used because the Bstate of CH3 is strongly predissociative, resulting innondetectable f luorescence quantum yields.

Approaching the sensitivity of laser-induced f luores-cence, degenerate four-wave mixing (DFWM) has beenshown in recent years to be a useful technique forthe detection of combustion-relevant species.11 In thisLetter we describe the use of a DFWM scheme forsensitively probing the concentration distribution ofCH3 radicals in a rich methane–air f lat f lame andcompare the results with absorption measurements.To our knowledge, this is the first example of thedetection of a polyatomic radical by DFWM.

The Herzberg b1 band near 216 nm is accessiblewith common laser sources and is what we study here.We used a tunable dye laser (Lambda Physik FL 2002)pumped with a frequency-doubled injection-seededNd:YAG laser (Positive Light) for excitation. WithDCM dye operation, the output was frequency doubledin a beta-barium metaborate sb-BaB2O4d crystaland then mixed with the dye-laser fundamental in asecond b-BaB2O4 crystal. The second-harmonic andfundamental radiation was then removed with mirrorscoated for 216 mm and a spatial filter (diamond pin-hole with a focusing telescope). Final output pulseshad energies of ,1 mJ near 216 nm and a bandwidth

0146-9592/95/192036-03$6.00/0

of , 0.2 cm21. For the DFWM measurements weused a nearly phase-conjugate geometry12 in whichthe forward pump and the probe beams are crossed ina horizontal plane and the backward pump beam isdirected downward at a small angle s,2±d with respectto the forward beams while it remains in a verticalplane containing the forward pump beam. Thiscauses the emerging signal to propagate downwardat the same angle, thus permitting very efficientcollection of the signal beam with a mirror rather thana beam splitter. Scattered light caused by the probebeam passing through the beam splitter is also avoided.For most of the experiments all beams had verticalpolarizations. Each beam had the same energy of10–20 mJypulse. The signal beam was directedthrough an , 2-mm iris, focused through a 25-mmpinhole, and detected with a UV-sensitive photomulti-plier tube (Hamamatsu 955) without any additionalfiltering. We typically averaged measurements from20 laser pulses, normalizing the DFWM signalsto I3 on a single-pulse basis, where I is the totalinput pulse energy. We simultaneously recordedabsorption data by measuring the transmitted pulseenergy of the probe beam and normalizing by I .(Optimized absorption measurements were performedseparately.) The signal-to-noise ratio in our experi-ments was 22 for the DFWM and 5 for the absorp-tion data.

All experiments used an axisymmetric honeycombburner (50-mm diameter) operated at atmosphericpressure with methane–air mixtures of f ­ 1.55.The f low conditions were chosen so that the f lamefront was nearly f lat and lifted, permitting access tothe region in which the peak CH3 concentration wasexpected based on calculations.13 The beam diametersused in the experiments ranged from 800 (FWHM) to200 mm, obtained by downcollimation with a telescopeand by additional insertion of 1-m focal-length lenses,respectively.

We used absorption spectroscopy to initially lo-cate and characterize CH3 distributions in the f lame.Tuning the laser frequency from 45 800 to 46 500 cm21,with the probe beam passing just below the visiblef lame zone, we obtained spectra such as that shown

1995 Optical Society of America

October 1, 1995 / Vol. 20, No. 19 / OPTICS LETTERS 2037

by the top trace of Fig. 1. (We found it necessaryto correct for slowly varying but significant changesin beam-splitter and mirror ref lectivities as a func-tion of frequency.) To confirm that the observedbroad absorption features are due to CH3, we calcu-lated the spectrum, using a symmetric top model asdescribed by Herzberg.14 Details for calculating theterm energies and the spectroscopic constants forthe ground state were found in Ref. 15. Since con-stants for the excited state are not well known,except for the electronic term energy and therotational constants,16 we varied the former for bestfit. With a predissociative broadening of 100 cm21

(Ref. 4) the calculated (solid curve at the top of Fig. 1)and measured spectra are in good agreement if astrong baseline absorption (dashed line) that increaseslinearly with frequency is added.4 Using a bandoscillator strength of 0.0137,17 we can reproduce theabsorption coefficient as measured by Davidson et al.5

Consequently, we calculated a peak CH3 concentrationof 650 parts in 106 occurring at a beam height of2.1 mm for our f lame, assuming a path length of 4 cm(confirmed with DFWM as described below).

Intense DFWM signals were observed when thebeam height was set to 2.1 mm and the laser wastuned to either of the CH3 absorption peaks. Froma comparison of the resulting DFWM spectra (bot-tom trace of Fig. 1) with the absorption results it isevident that the broad DFWM features originate fromCH3. The spectral positions of the two broad peaks,caused by the combined P and Q as well as theseparated R branch, are present in both spectra. This,together with our observation of the DFWM signalsat the f lame height corresponding to maximum CH3absorption, and in agreement with f lame-chemistrycalculations (discussed below), confirms that the de-tected signals originate from CH3. Interestingly, thestrong sloping background dominating the absorptionmeasurements is not present in the DFWM spectra.However, we did observe a nonresonant underlyingDFWM signal that was less than 10% of the peak CH3signal. Narrow lines are also observed in the DFWMsignal that are present in the absorption data to a muchlesser extent. All the narrow lines can be assigned toSchumann–Runge B –X transitions in molecular oxy-gen sO2d, mainly from v00 ­ 3–5.18 Differences in therelative intensities of spectral features between theabsorption and the DFWM spectra can be attributedto the factor a2sIyIsatd2 in the DFWM signal equa-tion, where a is the peak absorption coefficient.11 Forexample, we measured Isat for the DFWM backgroundto be roughly 50 times greater than that of CH3 orO2, explaining the relatively reduced background sig-nals in DFWM compared with absorption. Different(calculated) saturation intensities are also expected toreverse the DFWM peak intensities for the CH3 PyQand R branches compared with absorption.

An attempt to model the CH3 DFWM spectra with atwo-level model based on the theory of Abrams et al.19

and our theoretical absorption database did not yieldsatisfactory agreement with the measurements for theoverall shape of the spectrum. Calculating the spec-trum by simply squaring the calculated absorptionspectrum less the background yielded somewhat better

results (though judged too preliminary for publica-tion). We also observed that using a crossed (horizon-tal) polarization of the forward pump beam decreasedthe CH3 DFWM signal by a factor of 20 6 5. Bothof these results suggest a contribution from a ther-mal grating.20 A two-color experiment is currently inprogress to investigate this possibility.

We also investigated the applicability of DFWMfor real-time spatially resolved CH3 detection. Set-ting the laser frequency to the peak CH3 signalin a region free of O2 interference (46 185 cm21),we translated the burner vertically and observed adistinct peak in the signal at ,2.1 mm above theburner surface. Similar measurements with thelaser tuned away from resonance with CH3 and O2transitions indicated that the CH3 signal was thedominant contribution to the total signal by at leasta factor of 20. We found that the transverse spatialresolution defined by the focused laser beams easilyresolved the CH3 signal profile, which was asymmet-ric and approximately 1 mm thick. From geometricconsiderations we estimated the longitudinal resolu-tion to be 6.5 mm in this configuration. Thisestimate was consistent with comparisons of CH3profiles that we obtained by horizontally translatingthe burner first parallel and then perpendicular tothe laser beams. We found the CH3 signal strengthto be relatively constant—and to peak at the sameheight—across ,4 cm of the visible f lame diameterof 5 cm. Based on these results we used 4 cm asthe effective absorption path length for analyzing theabsorption measurements.

Figure 2 shows a CH3 spatial profile measured byDFWM in comparison with f lame-model calculations.The square root of the DFWM signal was taken follow-ing subtraction of the nonresonant background, but the

Fig. 1. Single-pass absorption spectrum of CH3 measured2.1 mm above the burner surface (top trace). A concentra-tion of 650 parts in 106 CH3 radicals is calculated for thisf lame position. The dashed line indicates the broadbandbackground absorption, dominating the overall absorptionin this spectral range. This background has been addedto the simulation of the spectrum (thick solid curve). Thebottom trace shows a DFWM spectrum of CH3 measuredwith 200-mm beams at the same position. The two under-lying broad features result from CH3, and the overlappingnarrow peaks correspond to O2 Schumann–Runge transi-tions. Note that there is region at 46 185 cm21 at the peakof the stronger CH3 band that is free of O2 interference.

2038 OPTICS LETTERS / Vol. 20, No. 19 / October 1, 1995

Fig. 2. Spatial CH3 profiles in a laminar methane–airf lame at f ­ 1.55. The measured profile (filled circles)was taken with 200-mm-diameter focused laser beams.

data have not been corrected for temperature effects,important only on the left side of the profile. Excel-lent agreement is observed for the shape of the profiles,with the peak concentration measured by DFWM beingroughly a factor of 2 lower (this discrepancy may resultfrom uncertainties in the absorption coefficient basedon the large background).

In comparison with other techniques for mea-suring CH3 we note the following advantages ofDFWM: Resonance-enhanced multiphoton ionizationmeasurements, which have been shown to be quitesensitive,8 suffer from the need to insert electrodesin the f lame, which can be intrusive or impracti-cal. Also, interferences that completely obscuredthe CH3 signals occurred in some regions of thef lame.8 Compared with line-of-sight methods (cavityring-down and conventional absorption techniques),DFWM provides spatial resolution in three dimen-sions, which is required in many applications. Theestimated CH3 detection limit with our current DFWMsetup is ,65 parts in 106, calibrated by absorptionmeasurements. This is comparable with 90 parts in106 for cw ring-laser absorption,5 assuming a detectionlimit of 0.2% for I0 and a path length equal to our inter-action length of ,6.5 mm. A detectivity of 16 partsin 106 for the same path length at 20-Torr totalpressure has been reported for cavity ring-downabsorption spectroscopy7; however, in atmospheric-pressure f lames this method has been demonstratedonly for detecting OH.21

Our results show that DFWM is a promising tech-nique for measuring CH3 concentrations with highspatial and temporal resolution and high sensitivity,even within 0.5 mm of surfaces. In contrast to thosewith absorption and resonance-enhanced multiphotonionization techniques, interfering background signalsare substantially reduced and are independent of fre-quency. Because of this, real-time measurements witha single laser wavelength appear possible, providedthat the temperature is known. The use of a standardtunable dye laser provides sufficient spectral resolu-tion to avoid simultaneous detection of O2. Futurestudy is aimed at developing a quantitative DFWMspectral model for CH3 and at investigating multicolorfour-wave mixing approaches.

The authors acknowledge helpful discussions withP. H. Paul of Sandia National Laboratories(SNL). This research was sponsored by the U.S.Department of Energy, Office of Basic EnergySciences, Chemical Sciences Division. V. Sick isgrateful to the Deutsche Forschungsgemeinschaft for agrant permitting the visit to SNL.

*Present address, Physikalish-Chemisches Institut,Universitat Heidelberg, Heidelberg, Germany.

References

1. J. Warnatz and C. Chevalier, ‘‘Reactions in the CyHyCsystem,’’ in Combustion Chemistry, W. C. Gardiner, Jr.,ed. (Springer-Verlag, New York, to be published).

2. S. J. Harris, Appl. Phys. Lett. 23, 2298 (1990).3. G. Herzberg and J. Shoosmith, Can. J. Phys. 34, 523

(1956).4. D. F. Davidson, A. Y. Chang, M. D. Di Rosa, and R. K.

Hanson, J. Quant. Spectrosc. Radiat. Transfer 49, 559(1993).

5. The reported value of kl was later revised. SeeD. F. Davidson, M. D. Di Rosa, E. J. Chang, andR. K. Hanson, J. Quant. Spectrosc. Radiat. Transfer53, 581 (1995).

6. T. Etzkorn, J. Fitzer, S. Muris, and J. Wolfrum, Chem.Phys. Lett. 208, 307 (1993).

7. P. Zalicki, Y. Ma, R. N. Zare, E. H. Wahl, J. R. Dadamio,T. G. Owano, and C. H. Kruger, Chem. Phys. Lett. 234,269 (1995).

8. K. C. Smyth and P. H. Taylor, Chem. Phys. Lett. 122,518 (1985).

9. P. B. Kelly and S. G. Westre, Chem. Phys. Lett. 151,253 (1988).

10. P. L. Holt, K. E. McCurdy, R. B. Weisman, J. S. Abrams,and P. S. Engel, J. Chem. Phys. 81, 3349 (1984).

11. R. L. Farrow and D. J. Rakestraw, Science 257, 1894(1992), and references therein.

12. R. L. Vander Wal, R. L. Farrow, and D. J. Rakestraw,in Twenty-Fourth Symposium (International) on Com-bustion (Combustion Institute, Pittsburgh, Pa., 1992),p. 1653.

13. M. N. Bui-Pham and J. A. Miller, in Twenty-FourthSymposium (International) on Combustion (Combus-tion Institute, Pittsburgh, Pa., 1994), p. 1309.

14. G. Herzberg, Molecular Spectra and Molecular Struc-ture III. Electronic Spectra of Polyatomic Molecules(Van Nostrand Reinhold, New York, 1966).

15. C. Yamada, E. Hirota, and K. Kawaguchi, J. Chem.Phys. 75, 5256 (1981).

16. K. Glanzer, M. Quack, and J. Troe, in Sixteenth Sym-posium (International) on Combustion (CombustionInstitute, Pittsburgh, Pa., 1977), p. 949.

17. A. B. Callear and M. P. Metcalfe, Chem. Phys. 14, 275(1976).

18. V. Sick and M. Szabadi, ‘‘Einstein coefficients for oxy-gen B –X transitions used in laser-induced f luores-cence experiments with tunable KrF excimer lasers,’’J. Quant. Spectrosc. Radiat. Transfer (to be published).

19. R. L. Abrams, J. F. Jam, R. C. Lind, and D. G.Steel, in Optical Phase Conjugation, R. A. Fischer, ed.(Academic, New York, 1983).

20. S. Williams, L. A. Rahn, P. H. Paul, J. W. Forsman, andR. N. Zare, Opt. Lett. 19, 1681 (1994).

21. G. Meijer, M. G. H. Boogaarts, R. T. Jongma, D. H.Parker, and A. M. Wodtke, Chem. Phys. Lett. 217, 112(1994).